MVSplat: Efficient 3D Gaussian Splatting from Sparse Multi-View Images

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Chen, Yuedong, Xu, Haofei, Zheng, Chuanxia, Zhuang, Bohan, Pollefeys, Marc, Geiger, Andreas, Cham, Tat-Jen, Cai, Jianfei
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866910669825114112
author Chen, Yuedong
Xu, Haofei
Zheng, Chuanxia
Zhuang, Bohan
Pollefeys, Marc
Geiger, Andreas
Cham, Tat-Jen
Cai, Jianfei
author_facet Chen, Yuedong
Xu, Haofei
Zheng, Chuanxia
Zhuang, Bohan
Pollefeys, Marc
Geiger, Andreas
Cham, Tat-Jen
Cai, Jianfei
contents We introduce MVSplat, an efficient model that, given sparse multi-view images as input, predicts clean feed-forward 3D Gaussians. To accurately localize the Gaussian centers, we build a cost volume representation via plane sweeping, where the cross-view feature similarities stored in the cost volume can provide valuable geometry cues to the estimation of depth. We also learn other Gaussian primitives' parameters jointly with the Gaussian centers while only relying on photometric supervision. We demonstrate the importance of the cost volume representation in learning feed-forward Gaussians via extensive experimental evaluations. On the large-scale RealEstate10K and ACID benchmarks, MVSplat achieves state-of-the-art performance with the fastest feed-forward inference speed (22~fps). More impressively, compared to the latest state-of-the-art method pixelSplat, MVSplat uses $10\times$ fewer parameters and infers more than $2\times$ faster while providing higher appearance and geometry quality as well as better cross-dataset generalization.
format Preprint
id arxiv_https___arxiv_org_abs_2403_14627
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle MVSplat: Efficient 3D Gaussian Splatting from Sparse Multi-View Images
Chen, Yuedong
Xu, Haofei
Zheng, Chuanxia
Zhuang, Bohan
Pollefeys, Marc
Geiger, Andreas
Cham, Tat-Jen
Cai, Jianfei
Computer Vision and Pattern Recognition
We introduce MVSplat, an efficient model that, given sparse multi-view images as input, predicts clean feed-forward 3D Gaussians. To accurately localize the Gaussian centers, we build a cost volume representation via plane sweeping, where the cross-view feature similarities stored in the cost volume can provide valuable geometry cues to the estimation of depth. We also learn other Gaussian primitives' parameters jointly with the Gaussian centers while only relying on photometric supervision. We demonstrate the importance of the cost volume representation in learning feed-forward Gaussians via extensive experimental evaluations. On the large-scale RealEstate10K and ACID benchmarks, MVSplat achieves state-of-the-art performance with the fastest feed-forward inference speed (22~fps). More impressively, compared to the latest state-of-the-art method pixelSplat, MVSplat uses $10\times$ fewer parameters and infers more than $2\times$ faster while providing higher appearance and geometry quality as well as better cross-dataset generalization.
title MVSplat: Efficient 3D Gaussian Splatting from Sparse Multi-View Images
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2403.14627